Wi-Fi 5 isn't slow. It's badly behaved in a crowd.
Every Wi-Fi generation up to Wi-Fi 5 solved the same problem harder: make one device go faster. None of them fixed what happens when forty devices want the air at once — which is every real home, office and shop floor. Wi-Fi 6 and Wi-Fi 7 finally did, by borrowing the scheduling and redundancy ideas that LTE and 5G proved at national scale.
The diagrams below animate. If your system is set to reduce motion, they hold a still frame instead.
Wi-Fi 5 makes devices queue. Politely, and very slowly.
Under 802.11ac and everything before it, the air is a single microphone in a room full of people. One device talks, everyone else waits. It works beautifully with three devices and falls apart with thirty — and the headline speed on the box tells you nothing about which situation you are in.
One talker at a time
Every device takes the whole channel for its turn, however small its packet. A smart plug sending 200 bytes occupies the same slot a 4K stream would.
Your neighbours make it worse
A legacy radio defers to any traffic it can hear, including the five networks in the next unit. In a flat or a shoplot row, much of your airtime is spent politely waiting for someone else's Netflix.
The slowest device sets the pace
An old phone at the edge of coverage transmits slowly, and while it does, nobody else gets the air. One weak client can drag a whole room down.
None of this is fixed by buying a faster Wi-Fi 5 router. The bottleneck is how the air is shared, not how fast one radio can go.
Four changes that rebuilt how the air is shared
Wi-Fi 6 is not a speed bump with a new number. It replaced "wait your turn" with "the access point schedules everyone", which is exactly how mobile networks have always worked.
The same three clients, served in a third of the airtime.
OFDMA — stop giving everyone the whole channel
The access point divides its channel into resource units and hands a slice to each client inside a single transmission. Small packets travel together instead of each one claiming the entire channel for a turn of its own.
The effect is not a bigger number on a speed test — it is that a busy room stops feeling busy. Latency under load falls, and the smart plugs stop stealing airtime from the video call.
BSS Coloring — perfect wireless harmony
Every network stamps its transmissions with a colour: a small identifier saying "this frame belongs to me". When your radio hears a frame, it checks the colour first. Same colour, it defers as it always did. Different colour, it recognises a neighbour and carries on transmitting.
That one change turns a corridor of competing networks into a set of networks that coexist. Nobody is shouting over anybody — each simply stops treating the others as its own traffic.
In flats, shoplot rows and dense offices this is usually the single biggest real-world gain in Wi-Fi 6 — bigger than any headline speed figure.
Same air, same instant. Your radio reads the colour, sees it is not its own, and keeps going.
MU-MIMO shapes independent beams, so three clients are served at once.
MU-MIMO — several conversations, one moment
Multiple antennas shape multiple beams, so the access point holds separate conversations with separate clients simultaneously. Wi-Fi 5 could do this downstream only; Wi-Fi 6 does it upstream too.
Combined with OFDMA, it means one slow device at the edge of coverage no longer freezes the room while it finishes talking.
Target Wake Time — scheduled naps
The access point and each device agree in advance when that device will wake up and transmit. In between, its radio genuinely sleeps rather than listening to everything on the channel.
Two wins at once: less airtime wasted on devices that had nothing to say, and dramatically better battery life on phones, tablets, cameras and battery sensors.
Radios sleep between their windows instead of listening to everything.
WPA3 comes with it
Wi-Fi 6 certification requires WPA3, which closes the offline dictionary attack that made weak WPA2 passwords guessable from a captured handshake. It also encrypts open networks individually, so a guest hotspot stops being a shared broadcast. If you are still on WPA2 with a memorable password, this alone is worth the upgrade.
Wi-Fi 6 fixed the scheduling. Wi-Fi 7 adds the redundancy.
If Wi-Fi 6 is about sharing the air fairly, Wi-Fi 7 is about never depending on one piece of it. The headline feature is the one network engineers recognise immediately.
Watch the middle link fail: the other two pick up the traffic without a reconnect.
MLO — link aggregation, over the air
If you have ever bonded two switch ports with 802.3ad / LACP, you already understand Multi-Link Operation. The client associates over 2.4, 5 and 6 GHz simultaneously and treats them as one logical connection.
- Aggregation
- Traffic spreads across the bands instead of committing to whichever one looked best at association time.
- Failover
- A microwave, a neighbour or a radar hit takes out one band — the others carry the session. No drop, no reconnect, no re-authentication.
- Low latency
- A frame can go out on whichever link is free right now rather than queueing behind the busy one.
- Low jitter
- Because stalls get routed around instead of waited out, the spread between best and worst packet narrows. Calls and remote desktops care about this far more than peak throughput.
This is the first time consumer wireless has had genuine link redundancy. It is the same instinct behind bonded uplinks and dual-homed switches, applied to the messiest medium in the building.
320 MHz channels — and the room to use them
Wi-Fi 7 doubles the widest channel to 320 MHz. That only helps where there is clean spectrum to spend, which is why it lives in the 6 GHz band — a band with no legacy devices, no microwave ovens and no twenty-year-old cordless phones.
Wider channels are the least interesting improvement in Wi-Fi 7, and the one most marketing leads with. Useful, but MLO is the reason to buy it.
Twice the width of Wi-Fi 6E — where the spectrum is quiet enough to matter.
More points per symbol means more bits — and less room for error.
4096-QAM — more bits in every symbol
A radio encodes data by placing each symbol at a point in a grid of amplitude and phase. Wi-Fi 5 used 256 points, Wi-Fi 6 pushed to 1024, and Wi-Fi 7 packs in 4096 — twelve bits in the time it used to send eight. Same airtime, half again as much data through it.
That extra headroom shows up in the jobs that move a lot of data over a short distance:
- Wireless VR and AR. A headset streaming rendered frames from a PC needs enormous, steady throughput; every millisecond saved in transfer is a millisecond off motion-to-photon latency.
- Working off a NAS. Editing video, pulling a RAW photo library or opening project files straight from shared storage without copying to the laptop first.
- Backups and machine images. The overnight jobs that only ran overnight because they were too slow to run any other time.
- High-bitrate local streaming. 8K and uncompressed sources to a TV or projector on the same floor.
- Bulk transfers at the desk. Dropping a day's footage off a camera or drone in minutes rather than over lunch.
All of that happens close to the access point — which is exactly where the tightest constellations hold. Treat 4096-QAM as a throughput feature for the room you are in, not a coverage one for the building.
Multi-RU and preamble puncturing — stop throwing the channel away
Interference rarely ruins a whole channel. Usually one 20 MHz slice goes bad — a neighbour, a radar hit, a badly behaved device — and under Wi-Fi 6 that is enough to force a fallback to a narrower channel. The healthy spectrum either side is simply abandoned.
Preamble puncturing cuts the bad slice out and keeps transmitting on everything else. Multi-RU then lets the access point hand those surviving slices out as separate resource units, to one client or several.
The result: a fault that used to cost you three quarters of your capacity now costs you one eighth of it.
One bad slice. Wi-Fi 6 abandons the channel; Wi-Fi 7 cuts round it.
Where each generation actually stands
Speeds quoted anywhere are theoretical maximums across all streams. Treat them as a ceiling nobody reaches, and read the right-hand column instead.
| Generation | Bands | Widest channel | What it changed |
|---|---|---|---|
| Wi-Fi 4 — 802.11n | 2.4 / 5 GHz | 40 MHz | Brought MIMO. Long obsolete; still the reason some networks crawl. |
| Wi-Fi 5 — 802.11ac | 5 GHz | 160 MHz | Faster for one device at a time. Downlink MU-MIMO only, and no answer to congestion. |
| Wi-Fi 6 — 802.11ax | 2.4 / 5 GHz | 160 MHz | OFDMA, uplink and downlink MU-MIMO, BSS Coloring, TWT, WPA3. The generation that fixed sharing. |
| Wi-Fi 6E | + 6 GHz | 160 MHz | Wi-Fi 6 with access to the clean 6 GHz band. No legacy devices to slow it down. |
| Wi-Fi 7 — 802.11be | 2.4 / 5 / 6 GHz | 320 MHz | MLO, 4096-QAM, Multi-RU, preamble puncturing. Redundancy and consistency. |
Both Wi-Fi 6 and Wi-Fi 7 are fully backward compatible. Old devices keep working — they just stop dictating how the rest of the network behaves.
What you will actually notice
Not "my speed test went up". These are the things people report after a proper Wi-Fi 6 or 7 deployment.
Calls stop breaking up
When someone else in the building starts a large download, your meeting carries on. Scheduled airtime instead of a free-for-all.
The evening slump disappears
In flats and terrace rows, the 8 p.m. slowdown is mostly your radio deferring to neighbours. BSS Coloring is what removes it.
Full rooms stay usable
A meeting room, classroom or cafe with forty devices behaves like a quiet one. This is the difference OFDMA makes.
Devices last longer per charge
Target Wake Time lets radios sleep properly. Phones, tablets and battery sensors all benefit without any configuration.
Roaming stops hurting
With Wi-Fi 7 and MLO, moving between areas no longer means a visible pause while a session re-establishes.
Fewer "the Wi-Fi is down" tickets
Most of those were never outages. They were congestion, and congestion is exactly what these generations address.
How to upgrade without wasting money
A new access point on a bad network is still a bad network. In roughly this order:
Check your devices
Features need both ends. MLO needs a Wi-Fi 7 client; 6 GHz needs a 6E or 7 client. We audit your actual device mix before recommending a generation.
Fix the backhaul first
An access point is only as good as the cable behind it. Wireless mesh throws away most of what Wi-Fi 6 gives you — hardwire it, or use FTTR where cable cannot go.
Survey, then place
Coverage, channel plan and power levels decided from measurements. Two well-placed APs beat five badly placed ones, at any generation.
Tune after occupancy
Re-check under real load with real people in the room. That is when BSS Coloring and OFDMA either prove themselves or reveal a placement problem.
Wi-Fi 6 and 7, answered
Want this designed properly for your building?
We survey first, tell you which generation your device mix actually justifies, and hardwire every access point. No guesswork, no upselling to the newest number.